VMD-L Mailing List

From: Joel Subach (mjsubach_at_alumni.ncsu.edu)
Date: Thu Apr 25 2024 - 09:39:46 CDT

Hi Josh thank you for your kind update and an enriching crystal clear
explanation:).

I will execute the default ffTK ! MP2 6-31G* TightSCF opt again and then
compare its CCGenFF Penalty Scores with the results of the RKS method and
Basis Set
and update you accordingly of the results (agreed that the MP2 will provide
greater CHARMM compatibility). Feel free to add any further comments.

Best,
Joel 🚀

On Thu, Apr 25, 2024 at 4:23 PM Josh Vermaas <vermaasj_at_msu.edu> wrote:

> I don't think "but someone on the internet said it was ok!" will fly in a
> methods section, so this is something that you'll need to figure out and be
> able to defend. What I will say is as far as I know the standard practice
> in CHARMM is HF or MP2, which is the FFTK default. RKS is a DFT method, and
> while DFT is often useful, it is sometimes almost too permissive of garbage
> inputs. If DFT isn't bailing, but the other methods are, are you sure you
> have the right charge and spin multiplicities set?
>
> -Josh
>
> On 4/25/24 6:42 AM, Joel Subach wrote:
>
> ...meaning also is the RKS Method acceptable considering it is not an MP2
> nor HF Method to my understanding? Thanks:)
>
> On Wed, Apr 24, 2024 at 4:00 PM Joel Subach <mjsubach_at_alumni.ncsu.edu>
> wrote:
>
>> Hello Josh thank you for your kind update:).
>>
>> I had amended the ! MP2 6-31G* TightSCF opt to ! RI-MP2 6-31G* TightSCF
>> opt autoaux rijcosx towards the Geometry Optimization and Charge
>> Optimization
>> which sped up the ORCA optimization results which successfully functioned
>> within Geom Opt and Charge Opt.
>>
>> I had then modified this same above Basis Set to the below in bold which
>> is thus far executing via ORCA without error(s) as-of-yet. RI-MP2 6-31G*
>> TightSCF opt
>> autoaux rijcosx, ! RI-MP2 def2-SVP/C TightSCF Opt NumFreq and ! RI-MP2
>> def2-TZVP def2-TZVP/C TightSCF Opt NumFreq PAL8, were generating errors
>> within ORCA, these three would of supposedly generate hessian output.out
>> files to then be used within the ffTK Bond and Angle Optimization.
>>
>> *! RKS PBE0 D3BJ def2-TZVP def2/J Opt Freq TightSCF RIJCOSX # (the
>> "def2/J" is the auxiliary coulomb fitting basis set)*
>>
>> Hopefully the above makes sense, my inquiry is if the above in bold and
>> the other Base Sets I used successfully in ORCA should be changed or do
>> these base
>> sets seem satisfactory to you.
>>
>> Thanks if you know:),
>> Joel 🚀
>>
>> On Wed, Apr 24, 2024 at 3:35 PM Vermaas, Josh <vermaasj_at_msu.edu> wrote:
>>
>>> Hi Joel,
>>>
>>> What basis set was successful? The rest of CHARMM36 uses HF/6-31G(d),
>>> and there are some other assumptions in parameterization that depend on
>>> this basis set. Like the multiplier by 1.16 between QM and MM energies is
>>> strictly empirical, and probably won't extend to other basis sets. If you
>>> are using some semi-empirical method, that probably won't be very
>>> compatible with the assumptions built into the CHARMM parameterization
>>> scheme, but if its still the HF method with a bigger basis set, that is
>>> probably fine.
>>>
>>> -Josh
>>>
>>> On 4/24/24 07:37, Joel Subach wrote:
>>>
>>> ....and the ffTK default ORCA Base Set is: ! MP2 6-31G* TightSCF opt
>>>
>>> On Wed, Apr 24, 2024 at 12:07 PM Joel Subach <mjsubach_at_alumni.ncsu.edu>
>>> wrote:
>>>
>>>> Hello VMD Forum,
>>>>
>>>> Towards the above topic, I have been executing the ffTK Tutorial
>>>> suggested QM Base Set, however, upon reaching the QM generated Hessian
>>>> Output the above base set has generated errors, accordingly I have instead
>>>> executed an ORCA Forum suggested Base Set which is so far executing
>>>> successfully.
>>>>
>>>> Would ffTK regardless suggest the ffTK Tutorial Base Set if functioning
>>>> in lieu of other Base sets since the HF/6-31G(d) level of theory is to
>>>> maintain the consistency with the CHARMM force field?
>>>>
>>>> Thanks if you know since if not to the above then I may go back and use
>>>> the ORCA Tutorial to modify each Base Set accordingly.
>>>>
>>>> Best,
>>>> Joel🚀
>>>>
>>>
>>> --
>>> Josh Vermaasvermaasj_at_msu.edu
>>> Assistant Professor, Plant Research Laboratory and Biochemistry and Molecular Biology
>>> Michigan State Universityvermaaslab.github.io <https://urldefense.com/v3/__http://vermaaslab.github.io__;!!HXCxUKc!ydv1Paa8DyAqa_YaWHP3rwEtzqGpWJ9L0W03Q-dUAqnQTzVQyqvDMLm76iu3CHvvN2chcAnXo79vyvedtpezU3A$>
>>>
>>>
> --
> Josh Vermaas
> vermaasj_at_msu.edu
> Assistant Professor, Plant Research Laboratory and Biochemistry and Molecular Biology
> Michigan State Universityvermaaslab.github.io
>
>